Yes, but by less than the headlines suggest, and the gap between the two commonly quoted numbers is the whole story. In running trials caffeine improved time to exhaustion by about 17%. In actual time trials — the thing that resembles a race — it improved performance by 0.71%. Both figures come from the same meta-analysis. The second one is the one to plan around.
How much faster does caffeine actually make you?
A 2022 meta-analysis pooled 21 randomised controlled trials covering 254 participants (167 recreational and 87 trained runners), with caffeine doses from 3 to 9 mg per kg of body mass, typically an hour before running.
- Time to exhaustion (7 studies, 12 comparisons): g = 0.392 (95% CI 0.214 to 0.571, p < 0.001) — a medium effect, averaging a 16.97% improvement.
- Time trial (14 studies, 22 comparisons): g = −0.101 (95% CI −0.190 to −0.012, p = 0.026) — a small effect, averaging 0.71% faster — though the reported variability (±0.83%) is larger than the average itself, so treat it as an order of magnitude rather than a precise figure.
Why the enormous gap? Because run-to-exhaustion tests exaggerate. Holding a fixed pace until you cannot is extremely sensitive to small changes in how hard something feels, so a modest physiological effect shows up as a huge percentage. It is the same statistical trap as the 32% sauna claim. When the same intervention is tested on a fixed distance, the effect shrinks to something realistic.
And that is not nothing. On a two-hour half marathon, 0.71% is about 51 seconds. Caffeine is one of the few legal aids with a genuine, replicated effect. It is just not the transformation it is sold as.
How much caffeine should you take, and when?
The range tested in the running trials was 3 to 9 mg/kg. A 2025 network meta-analysis of dose and delivery ranked low-dose capsules of around 3 mg/kg as the most effective strategy for time-trial performance, with moderate doses (4 to 6 mg/kg) and caffeine gum as viable alternatives. Two caveats on that ranking: the confidence intervals for low and moderate dose overlap heavily, so this is a ranking rather than a demonstrated superiority; and that review compared only low against moderate, so it says nothing about 7 to 9 mg/kg either way. It also drew on a broader endurance literature, much of it cycling. What it does support is that there is no established reason to go high.
For a 70 kg runner, 3 mg/kg is about 210 mg — roughly two cups of filter coffee. Note that a double espresso (about 125 mg) plus a cup of tea (about 45 mg) lands nearer 170 mg, and a cup of Indian milk tea is usually well under that, so do not assume you have hit the dose. Take it about 60 minutes before you start.
The honest caveats
The trials are overwhelmingly male. Of those 254 participants, 220 were men and 19 were women, with sex not reported for the remaining 15. Applying these numbers to women is an assumption, not a finding.
Caffeine is a diuretic argument that has been overstated, but in Indian summer heat, a stimulant that raises core temperature and heart rate is worth treating with more caution than a trial run at 18°C would suggest. Do not experiment with a new dose on a hot race day.
It disrupts sleep, and sleep is a bigger lever. Caffeine has a half-life of around five hours. An afternoon dose to get through a session can cost you the recovery that mattered more — see why sleep and recovery decide your running.
Never try it for the first time on race day. Some people get jittery, some get gut trouble. Test it in training.
References
Wang Z, Qiu B, Gao J, Del Coso J. Effects of caffeine intake on endurance running performance and time to exhaustion: a systematic review and meta-analysis. Nutrients, 2022;15(1):148. · Xue R, Huang J, Chen B, Ding L, Guo L, Cao Y, Girard O. Effects of caffeine dose and administration method on time-trial performance: a systematic review and network meta-analysis. Nutrients, 2025;17(23):3792. · Guest NS, et al. International Society of Sports Nutrition position stand: caffeine and exercise performance. J Int Soc Sports Nutr, 2021.